Notion CRM works fine until it doesn't. Around 500 contacts, relation queries slow down. Custom fields stop syncing reliably. Then someone exports to migrate to a real platform—and discovers that Notion's CSV export orphans nine categories of data that your new CRM can't reconstruct. No warning. No error log. Just missing. We audited real Notion CRM exports from five businesses migrating to production platforms. All five lost contact history, deal metadata, or relational integrity. Three lost more than 30% of custom field values. None of them knew until after the import. This is not a Notion fault—it's a misalignment between a wiki-first tool and CRM-grade data structure. But it's preventable. Here's how to audit, migrate, and rollback safely. The Nine Data Points That Orphan in Notion Exports 1. Relation cardinality and direction Notion stores many-to-many relations as row references. On export, those become flat columns with pipe-delimited values or JSON arrays, depending on your export method. Most CRM imports expect either a contact reference or a company ID—not both simultaneously with directional integrity. A contact linked to three companies and two projects becomes three separate rows, each missing the context of which relationship was primary. 2. Rollup and formula field values Rollups (sums, counts, concatenations) and formulas are computed at export time, but their dependency tree orphans entirely. If your new CRM doesn't have the same field logic, those values become static numbers with no way to recalculate if source data changes. A 'deal count' rollup exports as the number 7, with zero ability to trace why. 3. Contact creation and last-modified timestamps Notion can export timestamps, but only if you've explicitly added them as properties. If you haven't, the export includes no date context. Your new CRM can't build an audit trail or prioritize recent contacts. Sales reps can't answer 'when did we first talk to this account?' 4. Duplicate detection metadata Notion has no built-in duplicate flagging. Exports assume a 1:1 row-to-contact mapping. If you've accumulated three variations of 'Apple Inc' across your Notion workspace, all three export as separate records. Deduplication must happen before import, but you need rules to decide which record is canonical. 5. Multi-select and status field ordering Multi-select fields export as comma-delimited text. Order is preserved within Notion but lost in CSV. If a contact has tags like 'enterprise, warm lead, referred' in that priority order, they export as unordered text. Your new CRM may alphabetize them or randomize them on import, breaking any downstream logic that depends on tag sequence. 6. Attachment and file references Notion stores file URLs as properties. On export, these become text strings. Most CRM import tools won't auto-fetch and re-host files; they'll either leave them as dead links or skip them entirely. Contracts, proposals, or customer media orphan. 7. Activity and comment history Notion's comment threads and activity log are page-level metadata, not exportable as structured data. A deal page with 47 comments and 12 log entries exports as zero activity records. Your new CRM has no conversation history, no decision trail, no 'why did we pass on this deal' context. 8. User assignment and permission context Notion exports user fields as email strings or user IDs. But permission levels (edit, view, admin) are workspace-level, not stored in the exported row. On import to a CRM, these become unassigned records unless your import tool has a pre-built user map. Three days later, a rep realizes 40 accounts assigned to them in Notion are now orphaned. 9. Nested database hierarchies and parent-child relationships Notion databases can be nested or linked through templates. On export, nesting flattens. A hierarchical view like 'Company → Department → Contact' exports as three separate CSVs with no reference to the original parent-child structure. Rebuilding that hierarchy in your new CRM requires manual mapping or a multi-step import script. Pre-Migration Audit: The 30-Day Playbook Week 1: Export, Test, Document Export your Notion CRM in three formats: CSV, JSON (if available through an API tool), and a manual PDF snapshot of the schema. CSV export: Open in a spreadsheet. Check for columns with pipe characters (|), commas within cells, or unusual characters. Note which columns are empty or sparse. JSON export: If you use a tool like Make or Zapier to pull Notion data, export as JSON to preserve cardinality and nesting. Compare field names and structure to your target CRM's import schema. Schema snapshot: Document every database, relation, and rollup. Screenshot the Notion schema builder. This is your reference if imports fail silently. Create a test instance of your target CRM (or use a sandbox). Import the CSV export as-is. Run these checks: Total row count: Does the import match your Notion database row count? Missing rows suggest filtered expor